October 14 to November 25, 2026

Ottawa

This fall, the Ottawa Medical AI Research Institute (OMARI) is launching a new OMARI Seminar Series, bringing together researchers, clinicians, medical students, faculty and others interested in the rapidly evolving field of AI in medicine.

Starting on October 14, 2026, the seminars will take place in person every two weeks on Wednesday evenings (5 to 7 pm) at Roger Guindon Hall, room 2021. The series will explore a broad range of topics across medical AI: from traditional machine learning to large language models, from tabular and imaging data to text, and from model development and clinical evaluation to implementation and emerging methods and applications.

The series will begin as a four-session pilot, giving us an opportunity to bring the Ottawa medical AI community together and learn what topics, formats and discussions are most valuable to participants. If there is strong interest, we hope to build on these first sessions and develop the seminar into a continuing forum for exchanging ideas, sharing research and connecting people working at the intersection of AI and medicine.

Seminar I: October 14, 2026 | 5 to 7 pm

Inside ChatGPT: How Large Language Models Work, Learn, and Respond

Speaker: Dr. Fida Dankar

For the opening seminar, Prof. Khaled El Emam will introduce the series and Dr. Fida Dankar will join us as our first speaker. This talk provides an accessible introduction to how ChatGPT and other large language models work. It begins by placing generative AI within the broader history of artificial intelligence and introducing transformers. Using simple examples, it explains how text is converted into numerical representations, how attention uses surrounding words to understand meaning, and how models such as GPT generate text by repeatedly estimating probabilities and selecting the next token. It also explains how language models learn through large-scale pretraining, instruction fine-tuning, and human feedback. Finally, it introduces practical prompting methods and shows how clear instructions, reliable evidence, and careful checking can improve model responses. Register here.

Seminar II: October 28, 2026 | 5 to 7 pm

Beyond the AUC: Developing and Validating Clinical Prediction Models
A Case Study in Metabolic Gestational Age Estimation

Speaker: Dr. Steven Hawken

A model can have seemingly impressive AUC and other performance metrics and still give badly wrong individual-level estimates, fail in a new population, or offer little or no clinical benefit. This two-hour deep-dive shows how to develop and evaluate a prediction model properly and honestly and how to interpret model performance metrics reported in scientific papers.

The case study we will use is a set of machine learning models that estimate gestational age from routine newborn screening blood tests. These models were developed using Ontario data and externally validated internationally. Using the real models and data, the session shows why performance changes when a model travels, and what to do about it. It also includes a candid look at design choices and possible improvements with the benefit of hindsight. 

Participants get companion R code and open data to rerun every example. The session is suitable for researchers, trainees and clinicians comfortable with basic statistics; no engineering or coding background is needed. Register here.

Seminar III: November 11, 2026 | 5 to 7 pm

Using AI for Deductive Qualitative Data Analysis: From Practical Workflows to Questions of Quality and Rigour

Speakers: Dr. Arun Radhakrishnan, Dr. Zack van Allen and L. Jayne Beselt

This seminar explores how large language models can be used to support deductive qualitative data analysis in health research. Using a framework-based coding example, we will demonstrate practical approaches to developing prompts, structuring AI-assisted workflows, comparing model outputs, and reviewing AI-generated analyses. We will also discuss key considerations for responsible use, including data management, privacy, transparency, and verification of outputs.

Building on these practical examples, the session will explore emerging questions about how quality and rigour should be understood when AI is incorporated into qualitative analysis. What should we expect from a high-quality AI-assisted process? How should consistency, transparency, interpretation, and fidelity to the data be considered? And might those expectations differ between research and program evaluation? The session will combine hands-on learning with discussion of these evolving methodological questions. Register here.

Seminar IV: November 25, 2026 | 5 to 7 pm

From Pixels to Predictions: Developing AI Models for Medical Image Interpretation

Speakers: Dr. Kevin Dick and Youssef Megahed 

This hands-on workshop gives a technical overview of building computer vision models for medical imaging, from data preparation to clinical interpretation. Participants will explore key architectures, including convolutional neural networks, vision transformers and hybrid designs, and the choices that matter for medical data: scarce labels, class imbalance and domain shift. The session then turns to explainability, covering Class Activation Mapping (CAM) and its variants, with examples from ongoing work on MetaCAM++. Interactive demos run throughout on participants' own devices, connecting clinical intuition with model behaviour to support trustworthy, interpretable AI in medicine. Register here.

Registration

The link below is for the October 14 seminar only. Register for the next session today!